Authors: Low Tang Jung and Chan Piseth
Abstract: This paper presents a power management method based on system software process approach. This method is leveraging on the characteristics of the idle time of the applications running in a computer. It utilizes the applications idle time information to generate a timeout in controlling the computer display screen’s power state. The screen can be switched into the low power dimming state or standby state to save energy. The foreground applications’ idleness are monitored by a background monitoring process for predicting the next most likely timeout for each of the running applications. The prediction is through a simple moving average approach. The proposed method can also provide a clear view of unusual application idleness. The proposed method was tested in personal computers in a lab environment showing a 74 percent idle time reduction with an average of 85 percent prediction accuracy.
Keywords: power management; idle time profiling; moving average; power state; timeout prediction